Short Communication

Indirect impacts of invaders: A case study of the Pacific sheath-tailed bat(Emballonura semicaudata rotensis)

Jessica Nicole Welch ⁎, James A. Fordyce, Daniel S. SimberloffDepartment of Ecology and Evolutionary Biology, University of Tennessee, Knoxville, TN, USA

a b s t r a c ta r t i c l e i n f o

Article history:Received 9 September 2015Received in revised form 1 July 2016Accepted 4 July 2016Available online xxxx

Although many indirect consequences of biological invasions are plausible, few studies test hypotheses for man-agement of threatened taxa. A case study of the endangered Pacific sheath-tailed bat (Emballonura semicaudatarotensis) illustrates the importance of investigating indirect effects of invasion on species of conservation concern.Wehypothesized that two invaders, feral goats and Lantana camara, would indirectly affect the bat by decreasingavailability of suitable resources. Specifically, that microclimate and bat prey abundances in lantana shrub differfrom native forest habitat, and that preferential browsing by goats structures forests to be less suitable for bats.Our results suggest that bats avoid lantana shrub. However, we found no evidence that preferential goat brows-ing influenced bat activity. Our research implies that the impact of lantana on the persistence of the bat has beenunderestimated and that it is unclear how goats alter bat habitat aside from reducing understory vegetation. Fu-ture managers should prioritize efforts that restore native forest and reforest areas currently dominated by lan-tana. We urge conservation scientists to evaluate indirect effects of invasive species and publish findings thatelucidate the consequences for native populations.

© 2016 Published by Elsevier Ltd.

Keywords:Indirect effectsIsland invasionInvasion impactsLantana camaraFeral goatEmballonura semicaudata

1. Introduction

Global conservation for bats is needed: 15% of species are listed asextinct or threatened, 17% do not have a threat status owing to insuffi-cient data, and over half have unknown population trends (IUCN,2015). Moreover, bats are an oddity among mammals in that roughlya quarter of species are island endemics (Jones et al., 2009), makingthem important members of island communities. Given uncertainty re-garding population viability and their significant contribution to islanddiversity, research on threats to bats is a priority for conservation(Mickleburgh et al., 2002).

Endangerment and extinction occur disproportionately on islands,in part owing to island systems vulnerability to invaders (Gimeno etal., 2006). Biological invasion can have profound ecological and socio-economic impacts of isolated ecosystems (reviewed in Reaser et al.,2007). Studies on how biological invasions affect bats have focused ondirect interactions, such as predation and epidemiology of introducedpathogens (e.g., Fritts and Rodda, 1993; Rodriquez-Duran et al., 2010).Few have evaluated indirect effects — how interactions betweencoexisting species are affected by another species (Strauss, 1991) — onbats.

The Pacific sheath-tailed bat (Emballonura semicaudata) is a small in-sectivore designated “endangered” because its geographic area isb5000 km2 and highly fragmented (Bonaccorso and Allison, 2008).Emballonura semicaudata rotensis is a subspecies formerly distributedacross limestone islands of Guam and the Commonwealth of the North-ern Mariana Islands (CNMI) but now present only on the uninhabitedisland of Aguiguan (Wiles et al., 2011). The estimated 500 individualsof E. s. rotensis are commonly detected in forests and roost primarily inthree caves (Wiles et al., 2011). Consequently, E. s. rotensis is highly vul-nerable to stochastic demographic and environmental events.

Unconfirmed threats to E. s. rotensis include invasive species (Bergeret al., 2005). Shrub covers at least 20% of Aguiguan (150 ha) and is dom-inated by lantana (Lantana camara) (Amidon, 2009), a “top 100” inva-sive species (IUCN, 2001). Many studies have shown that lantana cangreatly alter native community structure (see review by Sharma et al.,2005), though its impacts on Aguiguan remain unstudied. Feral goats(Capra hircus) were introduced to Aguiguan during the mid-1800s(Butler, 1992), and their density has reached over 200 individuals persquare kilometer (Esselstyn et al., 2002), earning Aguiguan the nick-name “Goat Island”. The detrimental effects of goats on Aguiguan's na-tive species are largely anecdotal.

Here, we consider the conservation of E. s. rotensis to illustrate indi-rect impacts of invasion. We hypothesized that goat browsing and lan-tana spread indirectly impact E. s. rotensis by altering habitat structureand prey abundances. To evaluate our hypotheses, we ask: (1) Do batsbehave differently in lantana shrub than in native forest, and what

Biological Conservation 201 (2016) 146–151

⁎ Corresponding author at: 569 Dabney Hall, 1416 Circle Drive, Knoxville, TN 37996,USA.

E-mail address: jwelch14@utk.edu (J.N. Welch).

http://dx.doi.org/10.1016/j.biocon.2016.07.0040006-3207/© 2016 Published by Elsevier Ltd.

Contents lists available at ScienceDirect

Biological Conservation

j ourna l homepage: www.e lsev ie r .com/ locate /b ioc

factors account for this pattern? and (2) Does native understory coverindicate goat browsing, and is bat activity related to the structural com-plexity of native forest?

2. Methods

2.1. Study system

We conducted this study on Aguiguan, CNMI, (14°51′N, 145°33′E)fromMay 28 to June 14, 2013. The 7.09 km2 island consists of three con-centric limestone plateaus (Fig. 2). Native limestone forest occurs pri-marily along the smaller terraces of the island and on steep slopes.The dominant native tree species are Pisonia grandis, Cynometraramiflora, and Guamia mariannae (Esselstyn et al., 2002). Exotic grasses,Jatropha gossypiifolia, and Chromolaena odorata seedlings are inter-spersed with lantana across portions of the shrub habitat (hereafter,lantana shrub). Before the1930s Aguiguanwas coveredmostly in nativeforest, but lantana shrub is now extensive on the central plateau of theisland where vegetation was cleared during Japanese occupation(Butler, 1992). We performed all statistical analyses in the R statistical

computing environment (v3.0.3, R Development Core Team). The sup-plemental materials provide details of our study design.

2.2. Indirect effects of lantana

If lantana shrub provides an unfavorable microclimate for E. s.rotensis, lantana spread could indirectly affect bats as a consequence ofit replacing native forest (Fig. 1). Lantana shrub forms a densemonocul-ture with no canopy cover and little windbreak. Previous surveys onAguiguan detected E. s. rotensis in forests (Esselstyn et al., 2004;Gorresen et al., 2009),where themicroclimate is probablymore suitablefor a small bat (e.g., Rydell, 1991). Lantana shrub has likely limited theexpansion of native forest on Aguiguan because lantana is allelopathic(Gentle and Duggin, 1997), has bird-dispersed seeds (Turner andDowney, 2008), flourishes in disturbed environments (Duggin andGentle, 1998), and ungulates avoid its unpalatable foliage (Sharma etal., 2005). Moreover, patches of lantana expand within native forest asit has invaded treefall gaps opportunistically (Amidon et al., 2014).

To test if bat activity differed between native forest and lantanashrub, we recorded bat calls using acoustic detectors at six pairs of

Fig. 1.Hypotheses for the indirect effects of invaders, feral goats and Lantana camara, on Emballonura semicaudata rotensis. Dashed lines signify indirect interactions and solid lines signifydirect interactions. The arrow points to the resource user. The red path shows an interaction chain: lantana reduces native forest, which affects insect abundance and composition, whichaffects bat behavior. The blue path shows another interaction chain: lantana reduces native forest, which affects bat behavior. The green path shows exploitative competition: preferentialbrowsing by goats change the structure of native forest, which affects bat activity. (For interpretation of the references to color in thisfigure legend, the reader is referred to theweb versionof this article.)

147J.N. Welch et al. / Biological Conservation 201 (2016) 146–151

sites — each pair consisting of a forest and shrub habitat. Sites weremonitored for three consecutive nights. We determined bat activityusing the acoustic activity index (Miller, 2001), a tally of the numberof one-minute recording intervals per night containing two or moreconsecutive bat call pulses. We compared bat activity between nativeforest and lantana shrub using a hierarchical Bayesian model wherethe number of bat calls recorded in each habitat was modeled as beingdrawn from a binomial distribution described with parameter p. We as-sumed that the binomial parameters characterizing each pair of siteswere drawn from a beta distribution that represented island-level activ-ity between the two habitats (Fordyce et al., 2011). To determinewhichenvironmental factors correspondwith bat habitat use, we recorded theelevation of each site and its distance to the nearest of three caves hous-ing bats as well as nightly weather. We measured on-site temperatureat thirty-minute intervals, and we retrieved hourly wind speed(knots) and sky conditions (METAR) for each sampling night from theSaipan International Airport Weather Station, the nearest weather sta-tion recording nightly conditions.We calculated Spearman's rank corre-lation coefficient to examine the relationship between theseenvironmental factors and bat activity (R package “coin,” Hothorn etal., 2008).

Irrespective of howappropriate lantana shrubmicroclimate is, lanta-na spread could also indirectly affect bats if the habitat does not supportthe bat's insect prey (Fig. 1). Examination of E. s. rotensis fecal materialshowed that bats consume high volumes of wasps, moths, and beetles(Valdez et al., 2011). The interactions between lantana and Aguigan'snative insects have not been examined, but lantana flowers might at-tract bat prey, especially moths (Sharma et al., 2005). Conversely, proc-essed lantana has been used as a larvicide and insect repellant (Ogendoet al., 2006), so lantana shrub might not support every prey species.

To determine if abundances or composition of nocturnal, flying in-sects differed between native forests and lantana shrub, we sampled in-sects synchronically during bat acoustic monitoring using a miniatureblacklight trap and an aerial malaise trap with a bottom collector ap-proximately 50 m from each of the acoustic detectors. We identified in-sects at least to order and calculated abundance and richness for eachnight. For each sampling night and site, we summed insect abundanceby taxonomic order. We used distance-based redundancy analysis todetermine if insect composition differed between native forest and lan-tana shrub (Legendre and Anderson, 1999). We used a hierarchicalBayesianmodel to compare the abundance of each insect order between

native forest and lantana shrub, where insect abundance at each of thepaired sites was modeled as being drawn from a binomial distributionwith a beta prior (Fordyce et al., 2011). To determine how bat activitycorrespondswith insect abundance,we calculated Spearman's rank cor-relation coefficient for each pair of sites (R package “coin”, Hothorn etal., 2008). To elucidate if intrinsic habitat factors or insect availabilitydrives bat activity, we performed a redundancy analysis to relate bat be-havior, as measured by call characteristics (Table A4), to habitat and in-sect abundance (R package “vegan”, Oksanen et al., 2013).

2.3. Indirect effects of goats

Goats might indirectly affect bats by reducing native forest habitat,because goat browsing can impede regeneration of vegetation(Scowcroft and Hobdy, 1987) (Fig. 1). Goat numbers on Aguiguan arereportedly too high to allow seedling recruitment (e.g., Esselstyn et al.,2002; Amidon et al., 2014), although this conclusion is based on circum-stantial evidence. To determine if goats are responsible for low under-story recruitment on Aguiguan, we counted the number of seedlingsb1 m tall within a random 2 m diameter circle at ten locations inAguiguan forest. For comparison, we counted seedlings in native forestsof Tinian, a populated island approximately 8 km NE of Aguiguan thatshares a similar history of habitat degradation (Berger et al., 2005). Inparticular, we sampled in an area of native forest of approximately0.5 km2 that was historically browsed by goats and areas that had nohistory of goat browsing. We used a generalized linear model with aPoisson distribution to relate forest seedling counts to goat presence.

Goats might also indirectly affect bats through preferential brows-ing, which can affect habitat structure (Larkin et al., 2012). The propor-tional abundance of plant species can reflect their relative palatability toherbivores (Parsons et al., 1997), and changes in habitat structure owingto browsers can influence vertebrate species presence (e.g., Piana andMarsden, 2014). On Aguiguan, species of seedlings that appear fre-quently have been presumed unpalatable to goats (Esselstyn et al.,2002), but measurements of forest structure beyond height and trunkdiameter have not been reported. Habitat structure is relevant to man-aging E. s. rotensis because dense vegetation can hinder bat maneuver-ability (Muller et al., 2012) and prey detection (Arlettaz et al., 2001).E. s. rotensis wings are long and have above average aspect ratio andlow wing loading, which suggests slow hawking in open spaces(Norberg and Rayner, 1987). Therefore, bats might avoid areas within

Fig. 2.AmapofAguiguan Island, CNMI (Amidon, 2009).We illustrate three habitats: native forest (dark shade), non-native forest (mediumshade), lantana shrub (light shade). Thin,whitelines show topographic contour. Three white triangles denote the most important caves for bat roosting. White circles signify acoustic stations and insect sampling sites.

148 J.N. Welch et al. / Biological Conservation 201 (2016) 146–151

forests where goat browsing has favored growth of plants that createhigh structural complexity (hereafter, clutter) (Brigham et al., 1997).

To determine relative preferences of tree vegetation for goats, weperformed a cafeteria experiment (i.e., free-choice feeding) on 10 cap-tive goats from the feral population on Aguiguan using leaves of threedominant native forest tree species, P. grandis, C. ramiflora, and G.mariannae. We chose these species because a previous vegetation sur-vey (Esselstyn et al., 2002) suggested that C. ramiflora and G. mariannaedominate the small tree size classes because goats do not eat them andthat other species (e.g., P. grandis) have low regeneration because goatseat them. We determined the percentage of each species of leaves re-maining at the end of each trial, then coded the percentage as follows:4 = 75–100%, 3 = 50–75%, 2 = 25–50%, 1 = 0–25%. We used a hierar-chical Bayesianmodel to compare consumption of the three tree speciesby goats, where the percentage of each tree species leaves remainingafter each trial was modeled as a multinomial distribution describedwith a theta prior (Fordyce et al., 2011).

To determine if habitat structure is related to bat activity, we record-ed vegetation characteristics at each native forest sitewithin 15mof theacoustic detector. In particular, we used measurements of P. grandis, C.ramiflora, and G. mariannae (the three species used in the cafeteria ex-periment above) to estimate relative amounts of clutter at three-meter height intervals. Because these trees vary in densities andbranching heights, we expected that bat activity would reflect clutteramount owing to preferential goat browsing. In particular, we predictedthat bat activity would be low at locations with high amounts of G.mariannae becausewe have observed that this species creates high clut-ter in the forest understory relative to the other two tree species. Wecalculated Spearman's rank correlation coefficient to determine the re-lationship between bat activity and habitat structure (Table A7) (Rpackage “coin”, Hothorn et al., 2008).

3. Results & discussion

3.1. Do bats behave differently in lantana shrub than in native forest, andwhat factors account for this pattern?

Bats were more active in native forest than lantana shrub (probabil-ity of recorded calls in native forest N lantana shrub = 0.993; Table A2;Fig. A1), though bats were not detected at every native forest site (Fig.A2). Considerable differences in bat activity across native forest sitessupport the hypothesis that within-habitat structure is an importantconsideration for species management (Sharma et al., 2005). Batswere relatively inactive in lantana shrub when temperature was high(ρ = −0.198, P = 0.009; Table A3) and wind was strong(ρ = −0.199, P = 0.006). Strong winds likely inhibit bat navigation inlantana shrub, and inactivity during the high temperatures (relative tonightly averages) at dusk and dawn likely indicates retreat into thecover of forest. Given this interpretation, microclimate seems a reason-able explanation for low bat activity in lantana shrub. Contrarily toGorresen et al. (2009), we found no strong correlation between bat ac-tivity and distance to known roosts (ρ=−0.007, P=0.993).Moreover,elevationwas not related to activity (ρ=−0.282, P=0.369),which in-dicates bats are not limited in their distribution across Aguiguan.

Overall insect abundance was higher in native forest than lantanashrub (P(native forest N lantana shrub) N 0.99; Table A2; Fig. A3), andhabitat explained 29% of variation in insect community composition(F1,16 = 4.049, P = 0.015). Given their variety of insect prey (Valdezet al., 2011), we assume that E. s. rotensis will pursue flying insectsthat are the appropriate size for the bat to handle. Median body lengthof all insects captured was 4.5 mm (range of 1–16 mm), likely the sizeof insects consumed by E. s. rotensis given its b7 g body weight (Wileset al., 2011).

There were more dipterans, hymenopterans, microlepidopterans,and neuropterans in native forest (all P(native forest N lantanashrub) N 0.900, respectively; Table A2), but more isopterans in lantana

shrub (P(native forest b lantana shrub) = 0.933). High abundances ofhymenopterans and microlepidopterans in native forest, where batsare most active, concur with an earlier study showing that these insectsmake up a large volume of the bat's diet (Valdez et al., 2011). Althoughonly a small percentage of their diet consisted of dipterans, it is possiblethat bats eat a larger volume of these, but soft bodies are not easily locat-ed within the bats' feces (Dickman and Huang, 1988). There was no dif-ference between habitats in beetle numbers though theywere frequentin bat fecal pellets. Additional studies on the diet of bats, seasonal vari-ation of insect abundances, and plant-insect interactions will elucidatepossible indirect impacts of changes in forestation on bats via declinesof tree species important to the bat's particular phytophagous prey.

We recorded 83 feeding buzzes (i.e., a rapid number of call pulseswithin a short period that indicates a bat encountering prey; Griffin etal., 1960), 92% ofwhichwere in native forest.We found amoderate pos-itive correlation between bat activity and insect abundance(ρ = −0.566, P b 0.001). Forty-five percent of variation in bat callswas explained by habitat (F1,16 = 15.269, P = 0.001; Table A4; Fig.A4) with call duration correlating positively with lantana shrub (Fig.A5), but we failed to detect a relationship between call parametersand insect abundance (R2 = 0.08, F1,16 = 2.867, P = 0.093). We inter-pret these results to suggest that habitat is a stronger driver of E. s.rotensis behavior than is insect abundance. Accordingly, the correlationbetween bat activity and insect abundance might reflect a correspond-ing reliance on native forest microclimate (Peng et al., 1992).

3.2. Does native understory cover indicate goat browsing, and is bat activityrelated to the structural complexity of native forest?

Our results show that the understory of Aguiguanhas very few seed-lings compared to both historically browsed and unbrowsed native for-ests on Tinian (Χ 2=352.38, df=1, P ≤ 0.001; Table A5),which suggeststhat goats limit regeneration of vegetation on Aguiguan. Although wedid not confirm the species of seedlings, previous studies show thatmost seedlings in Aguiguan's native forest are G. mariannae and C.ramiflora (Esselstyn et al., 2002), curiously, G. mariannae has very highand P. grandis has very low densities in native forest understory acrossGuam and the CNMI regardless of browsing history (Mueller-Domboisand Fosberg, 1998).Moreover, anecdote and aerial photographs suggestthat the perimeter of native forest has not receded despite heavy under-story browsing by goats.

Goats strongly preferred P. grandis and C. ramiflora foliage (low clut-ter tree species) over that of G. mariannae (high clutter tree species)(P(P. grandis N G. mariannae) N 0.99, P(C. ramiflora N G. mariannae) =0.997, P(P. grandis N C. ramiflora)=0.76; Table A6). Bat activity correlat-ed negatively with total stem count (ρ=−0.812, P= 0.073) and pos-itively with average canopy height (ρ = 0.998, P = 0.006), but therewas no relation to tree species except for one anomaly (C. ramifloracrown volume 3–6 m (ρ=−0.841, P= 0.043); Table A7). This is con-sistent with our hypothesis that bats are more active in less clutteredforests (Gorresen et al., 2009), but not that they are affected by the clut-ter of particular tree species. The lack of an effect suggests that otherprocesses are the main determinants of habitat structure on Aguiguan,or that there is a complicated interplay between goat browsing and nat-ural disturbance on forest dynamics (e.g., Rojas-Sandoval et al., 2014).Nevertheless, because E. s. rotensis forages in spaces with low clutter,processes that reduce availability of open forest decrease the amountof suitable bat foraging habitat.

4. Conclusion

To improve management of endangered species, a first step is toidentify and alleviate ongoing threats. Although indirect impacts of in-vasive species on native populations are plausible in many cases, mostproposed consequences of invasion lack supporting research. Our re-sults indicate that indirect impacts of lantana on the population of E. s.

149J.N. Welch et al. / Biological Conservation 201 (2016) 146–151

rotensis have been underestimated. Bats are largely inactive in lantanashrub, where the microclimate is harsh and prey are scarce, making re-maining tracts of Aguigua's native forest a vital resource. Although goatscan be strongdrivers of habitat change, it is unlikely that browsing alonewill cause native forest to disappear. By contrast, if lantana impedesgrowth of native trees, the extent of forested habitat on Aguiguanmight gradually recede, making lantana spread a concern beyond thecontext of E. s. rotensis conservation. Because the range of E. s. rotensisis currently limited to Aguiguan, declines of the population's carryingcapacity owing to a lack of suitable resourcesmight lead to demograph-ic and genetic problems associated with small population size (e.g.,Shaffer, 1981).

Conservation assessments for E. s. rotensis advise that goats shouldbe eradicated in order to conserve bat habitat (Berger et al., 2005).Still, it is unclear how goats alter Aguiguan's native forest aside from re-ducing understory vegetation, because forested habitat remains despitehigh goat densities over the past 200 years. Aguiguan's jurisdictive au-thority strongly enforces regulations to preserve the goat population,making permanent removal of goats unlikely without substantial evi-dence of harm to endangered fauna. No management strategies havebeen implemented for removing or managing lantana on Aguiguan,even though lantana is considered a serious pest in similar environ-ments (Day et al., 2003).We support removal of goats, as thiswould un-doubtedly lead to an increase in native vegetation (e.g., Kessler, 2011),but we insist that lantana should also be managed and subsequent re-forestation should be attempted in order to restore E. s. rotensis habitaton Aguiguan.

Though the notion is untested, goats might negatively affect nativevegetation and facilitate lantana spread because goats do not browselantana (“indirect mutualism” in Wootton, 1994). In fact, a study onplant community succession found that lantana persisted longer incleared plots with subsequent goat browsing than without (Larkin etal., 2012). We recommend long-term goat exclusion plots to character-ize the regrowth of vegetation, and to examine if regeneration of nativespecies is possible in lantana shrub if goats are absent. Additionally,wildlife managers should limit disturbances that might favor lantanaspread. Multiple invaders within a community can affect the assem-blage of native communities and alter interactions amongnative species(Kuebbing et al., 2013). Investigating interactions between common in-vasives species, like goats and lantana, will undoubtedly have broad rel-evance to invasion biology and conservation management.

We have found that indirect effects of an exotic plant and vertebrateherbivore might jeopardize long-term persistence of an endemic, en-dangered bat. The ubiquity of invasion by feral goats and Lantanacamara makes our study relevant to conservation biology worldwide,and adds to the burgeoning topic of indirect effects. More generally,we contend that scientists should more frequently examine indirect ef-fects of invasive species on threatened taxa.

Acknowledgments

We thank the following for their expertise and logistical assistanceon this project: Gary McCracken, Tammy Mildenstein, Haldre Rodgers,Ton Castro, Anthony Deleon Gurrero, Commonwealth of the NorthernMariana Islands Department of Fish andWildlife, Mayor's Office of Tini-an. We thank Zach Marion, St. Thomas LeDoux, and anonymous re-viewers for assistance with the manuscript. Funding: This work wassupported by Sigma Xi [grant number G20120315161439]; Bat Conser-vation International [grant number A15-0240-001].

Appendix A. Supplementary data

Supplementary data to this article can be found online at http://dx.doi.org/10.1016/j.biocon.2016.07.004.

References

Amidon, F., 2009. Vegetation Surveys on Tinian and Aguiguan in Terrestrial Resource Sur-veys of Tinian and Aguiguan. Mariana Islands. US Fish and Wildlife Service, Hawaii,pp. 8–18.

Amidon, F., Camp, R.J., Marshall, A.P., Pratt, T.K., Williams, L., Radley, P., Cruz, J.B., 2014.Terrestrial Bird Population Trends on Aguiguan (Goat Island), Mariana Islands. BirdConservation International, New York.

Arlettaz, R., Jones, G., Racey, P.A., 2001. Effect of acoustic clutter on prey detection by bats.Nature 414, 742–745.

Berger, G.M., Gourley, J., Schroer, G., 2005. Comprehensive wildlife conservation strategyfor the CNMI. Technical Report, CNMI-DFW.

Bonaccorso, F., Allison, A., 2008. Emballonura semicaudata. IUCN red list of threatened spe-cies (Version 2011.1).

Brigham, R.M., Grindal, S.D., Firman, M.C., Morissette, J.L., 1997. The influence of structuralclutter on activity patterns of insectivorous bats. Can. J. Zool. 75, 131–136.

Butler, B.M., 1992. An Archaeological Survey of Aguiguan (Aguijan), Mariana Islands. Mi-cronesian Archaeological Survey Report No. 29. Division of Historic Preservation, Sai-pan, p. 260.

Day, M.D., Wiley, C.J., Playford, J., Zalucki, M.P., 2003. Lantana: Current Management Sta-tus and Future Prospects. ACIAR Monograph, Canberra, p. 102.

Dickman, C.R., Huang, C., 1988. The reliability of fecal analysis as a method for determin-ing the diet of insectivorous mammals. J. Mammal. 69, 108–113.

Duggin, J.A., Gentle, C.B., 1998. Experimental evidence on the importance of disturbanceintensity for invasion of Lantana camara L. In dry rainforest-open forest ecotones innortheastern NSW, Australia. For. Ecol. Manag. 109, 279–292.

Esselstyn, J.A., Cruz, J.B., Williams, L.L., Hawley, N., 2002. Technical report 9. CNMI Divisionof Fish and Wildlife, Saipan, p. 55.

Esselstyn, J.A., Wiles, G.J., Amar, A., 2004. Habitat use of the Pacific sheath-tailed bat(Emballonura semicaudata) on Aguiguan, Mariana Islands. Acta Chiropterol. 6,303–308.

Fordyce, J.A., Gompert, Z., Forister, M.L., Nice, C.C., 2011. A hierarchical Bayesian approachto ecological count data: a flexible tool for ecologists. PLoS One 6.

Fritts, T.H., Rodda, G.H., 1993. The role of introduced species in the degradation of islandecosystems: a case history of Guam. Annu. Rev. Ecol. Syst. 29, 113–140.

Gentle, C.B., Duggin, J.A., 1997. Allelopathy as a competitive strategy in persistentthickets of Lantana camara L. In three Australian forest communities. PlantEcol. 132, 85–95.

Gimeno, I., Vila, M., Hulme, P.E., 2006. Are islands more susceptible to plant invasion thancontinents? A test using Oxalis pes-caprae L. In the western Mediterranean.J. Biogeogr. 33, 1559–1565.

Gorresen, P.M., Bonaccorso, F.J., Pinzari, C.A., 2009. Habitat occupancy and detection of thePacific sheath-tailed bat (Emballonura semicaudata) on Aguiguan, Commonwealth ofthe Northern Mariana Islands. Acta Chiropterol. 11, 331–342.

Griffin, D.R., Webster, F.A., Michael, C.R., 1960. The echolocation of flying insects by bats.Anim. Behav. 8, 141–154.

Hothorn, T., Hornik, K., van deWiel, M.A., Zeileis, A., 2008. Implementing a class of permu-tation tests: the coin package. J. Stat. Softw. 28, 1–23.

IUCN, 2001. Invasive species specialist group. 100 World's worst invasive alien species.http://www.issg.org/booklet.pdf.

IUCN, 2015. The IUCN red list of threatened species. Version 2015-4. bhttp://www.iucnredlist.orgN.

Jones, K.E., Mickleburgh, S.P., Sechrest, W., Walsh, A.L., 2009. Global overview of the con-servation of island bats: importance, challenges, and opportunities. In: Fleming, T.H.,Racey, P.A. (Eds.), Island Bats. Chicago Press, Chicago 496–230.

Kessler, C.C., 2011. Invasive species removal and ecosystem recovery in the MarianaIslands; challenges and outcomes on Sarigan and Anatahan. In: Veitch, C.R., Clout,M.N., Towns, D.R. (Eds.), Island Invasives: Eradication and bAnagement. IUCN, Swit-zerland, pp. 320–324.

Kuebbing, S.E., Nunez, M.A., Simberloff, D., 2013. Current mismatch between research andconservation efforts: the need to study co-occurring invasive plant species. Biol.Conserv. 160, 121–129.

Larkin, C.C., Kwit, D., Wunderle Jr., J.M., Helmer, E.H., Stevens, M.H.H., Roberts, M.T.K.,Ewert, D.N., 2012. Disturbance type and plant successional communities in Bahamiandry forests. Biotropica 44, 10–18.

Legendre, P., Anderson, M.J., 1999. Distance-based redundancy analysis: testing multispe-cies responses in multifactorial ecological experiments. Ecol. Monogr. 69, 1–24.

Mickleburgh, S.P., Hutson, A.M., Racey, P.A., 2002. A review of the global conservation sta-tus of bats. Oryx 36, 18–34.

Miller, B.W., 2001. A method for determining relative activity of free flying bats using anew activity index for acoustic monitoring. Acta Chiropterol. 3, 93–105.

Mueller-Dombois, D., Fosberg, R.F., 1998. Vegetation of the Tropical Pacific Islands. Spring-er, New York, p. 881.

Muller, J., Mehr, M., Bassler, C., Fenton, M.B., Hothorn, T., Pretzsch, H., Klemmt, H.J., Brandl,R., 2012. Aggregative response in bats: prey abundance versus habitat. Oecologia 169,673–684.

Norberg, U.M., Rayner, J.M.V., 1987. Ecological morphology and flight in bats (Mammalia;Chiroptera): wing adaptations, flight performance, foraging strategy and echoloca-tion. Philos. Trans. R. Soc. Lond. Ser. B Biol. Sci. 316, 335–427.

Ogendo, J.O., Belmain, S.R., Deng, A.L., Walker, D.J., 2006. Comparison of toxic and repel-lent effects of Lantana camara L. with Tephrosia vogelii and a synthetic pesticideagainst Sitophilius zeamais Motschulsky (Coleoptera: Curculionidae) in stored maizegrain. Insect Sci. Appl. 23, 127–135.

Oksanen, J., Blanchet, F.G., Kindt, R., Legendre, P., Minchin, P.R., O'Hara, R.B., Simpson, G.L.,Solymos, P., Stevens, M.H.H., Wagner, H., 2013. Vegan: community ecology package.R package version 2.0–10. http://CRAN.R-project.org/package=vegan.

150 J.N. Welch et al. / Biological Conservation 201 (2016) 146–151

Parsons, D.A.B., Shackleton, C.M., Scholes, R.J., 1997. Changes in herbaceous layer condi-tion under contrasting land use systems in the semi-arid lowveld, South Africa.J. Arid Environ. 37, 319–329.

Peng, R.K., Fletcher, C.R., Sutton, S.L., 1992. The effect of microclimate on flying dipterans.Int. J. Biometeorol. 36, 69–76.

Piana, R.P., Marsden, S.J., 2014. Impacts of cattle grazing on forest structure and raptor dis-tribution within a neotropical protected area. Biodivers. Conserv. 23, 559–572.

Reaser, J.K., et al., 2007. Ecological and socioeconomic impacts of invasive alien species inisland ecosystems. Environ. Conserv. 34, 98–111.

Rodriquez-Duran, A., Perez, J., Montlban, M.A., Sandoval, J.M., 2010. Predation by free-roaming cats on an insular population of bats. Acta Chiropterol. 12, 359–362.

Rojas-Sandoval, J., Meléndez-Ackerman, E.J., Fumero-Cabán, J., García-Bermúdez, M.A.,Sustache, J., Aragón, S., Morales, M., Fernández, D.S., 2014. Effects of hurricane distur-bance and feral goat herbivory on the structure of a Caribbean dry forest. J. Veg. Sci.25, 1069–1077.

Rydell, J., 1991. Seasonal use of illuminated areas by foraging northern bats Eptesicusnilssoni. Ecography 14, 203–207.

Scowcroft, P.G., Hobdy, R., 1987. Recovery of goat-damaged vegetation in an insular trop-ical montane forest. Biotropica 19, 208–215.

Shaffer, M.L., 1981. Minimum population sizes for species conservation. Bioscience 31,131–134.

Sharma, G.P., Raghubanshi, A.S., Singh, J.S., 2005. Lantana invasion: an overview. WeedBiol. Manage. 5, 157–165.

Strauss, S., 1991. Indirect effects in community ecology: their definition, study and impor-tance. TREE 6, 206–210.

Turner, P.J., Downey, P.O., 2008. The role of native birds in weed invasion, species decline,revegetation and reinvasion. Consequences for Lantana Management in Proc. 16thAustralian Weeds Conf. Brisbane Queensland Weeds Society, pp. 30–32.

Valdez, E.W., Wiles, G.J., O'Shea, T.J., 2011. Diets of the sympatric Pacific sheath-tailed bat(Emballonura semicaudata rotensis) and Mariana swiftlet (Aerodramus bartscht) onAguiguan, Mariana Islands. Pac. Sci. 65, 301–309.

Wiles, G.J., O'Shea, T.J.,Worthington, D.J., Esselstyn, J.A., Valdez, E.W., 2011. Status and nat-ural history of Emballonura semicaudata rotensis on Aguiguan, Mariana Islands. ActaChiropterol. 13, 299–309.

Wootton, J.T., 1994. The nature and consequences of indirect effects in ecological commu-nities. Annu. Rev. Ecol. Syst. 25, 443–466.

151J.N. Welch et al. / Biological Conservation 201 (2016) 146–151